1. Introduction

CO₂ capture from flue gases is now vital to reduce the green-house effect. As a main process to remove CO₂ from the industrially emitted gases, the chemical absorption to aqueous solutions of low molecular weight amines, such as monoethanolamine (MEA) and N-methyldiethanolamine (MDEA), has been traditionally utilized. However, this conventional procedure has some serious drawbacks, e.g., inevitable loss and emission of those corrosive amines and thermal degradation during the energy-demanding regeneration process of the absorbents [1] . As a promising candidate for CO₂-absorbing materials, amine-func- tionalized ionic liquids have recently attracted much attention [1] - [10] . Since ionic liquids (ILs) have favorable properties such as negligible volatility and excellent thermal stability, they are free from some of the above drawbacks. A prominent example for the application of ILs for the CO₂ absorption has been reported by Romanos et al. [11] who employed the “inverse supported ionic liquid phase materials” (inverse SILPs). SILP is a material in which ILs are supported or adsorbed on the surface of porous substrate. In the “inverse” SILP, which is a powder-like substance, the phases are inversed and the IL is surrounded by solid materials. For example, the inversed SILP was prepared by a two-step method; in the first step an amino acid-derived IL such as N, N, N-tri- methyl-N-ethyl ammonium propionate was mixed with fumed silica nanoparticles. Then the mixture was slurried with ethanol and the suspension thus obtained was dried to obtain the final powdery substance. This substance may be called Dry Ionic Liquid (D-IL) according to our original naming [12] . However, the D-IL can be prepared, in principle, with a much easier procedure, i.e., by simply mechanically mixing an appropriate IL with a hydrophobized fumed silica.

In the present study we prepared aqIL-DMs by using some amino acid-based ILs, because their CO₂ absorption abilities may be rather enhanced in the presence of water [16] . The performance as CO₂-absorption materials was checked in terms of the absorption speed and the capacity. Since water absorbs CO₂ only slightly, the addition of water to ILs would reduce the total mass-base CO₂ absorption ability even if the mol-base ability of IL employed were enhanced by the presence of water. Thus, we also tested an aqIL-DM in which water was replaced by aqueous solution of poly(allyl amine) (PAlAm) because the polyamine should have a substantial CO₂ absorption ability [13] [17] . The resultant total CO₂ absorption ability, in fact, proved to become higher than either of the respective bulk absorptions.

2. Materials and Methods

2.1. Materials

Tetraethyl ammonium hydroxide ([N₂₂₂₂][OH]) and three kinds of amino acids, i.e., glycine [Gly], L-alanine [Ala] and L-lysine [Lys], were purchased from Sigma Aldrich as 35 wt% aqueous solution and powder samples, respectively, and used as received. Poly(allyl amine)hydrochloride (PAlAm∙HCl) (MW:15000, Sigma Aldrich) was neutralized by NaOH and then filtrated through an Ultrafilter UP-10 (Advantec) for several times with pure water to remove inorganic ions and low molecular weight polymers. Concentration of PAlAm thus purified was estimated by conductometric titration to be ca. 24 wt%.

Hydrophobic fumed silica particle (HDK-H18, primary particle size: 5 - 30 nm), the surface OH groups of which are methylated with poly(dimethyl siloxane) by 75%, was purchased from Wacker Asahikasei Silicone Co. and used as received.

2.2. Methods

2.2.1. Preparation of Amino Acid-Based AqILs

[N₂₂₂₂][Gly], [N₂₂₂₂][Ala], and [N₂₂₂₂][Lys] ILs were prepared by neutralizing the respective amino acids with the equimolar [N₂₂₂₂][OH]. For example, in the case of [N₂₂₂₂][Gly] 1.0 g of glycine was neutralized with 5.6 g of the 35 wt% of [N₂₂₂₂] [OH] solution to obtain 6.6 g of [N₂₂₂₂][Gly] containing 55.2 wt% of water (i.e., 45 wt% aqueous [N₂₂₂₂][Gly] solution). In a similar way we prepared [N₂₂₂₂] [Ala] and [N₂₂₂₂][Lys] ILs containing 54 wt% and 48 wt% of water, respectively.

2.2.2. Preparation of D-IL

AqIL-DMs were prepared with a blender machine (Waring J-SPEC blender, container volume: 50 mL) by mixing 6.0 g of an aqueous IL solution and 1.05 g of the silica at a fixed speed (22,500 rpm) for 90 s (30 s × 3 with an interval of 10 s each) at a room temperature. Partial remove of water from the prepared DM samples was performed with an infrared humidity meter (FD-720, Kett Co. Ltd.) at 40˚C or 80˚C. AqIL-DM containing PAlAm was also prepared with the same preparation procedure except for using mixed solutions of aq. ILs and aq. PAlAm solution (9.9 wt%).

2.2.3. CO₂ Absorption Measurements

Absorption performance of aqIL-DM and bulk liquids (aq. solution of IL, aq. solution of PAlAm and their mixture) for CO₂ was estimated by measuring pressure depression within connected glass cylinders (Hyper Glass Cylinder HPG-96-3 (90 mL), Taiatsu Garasu Kogyo Co. Ltd.) due to the absorption of CO₂ into 2 g of DM or bulk sample. The experimental set up which was assembled within a thermostat incubator (FMU-133I, Fukushima Kogyo Co. Ltd.) and the detailed description on the measurements were given in our previous report [13] . The absorption ability (A.A.) was measured or expressed by three kinds of ways; mol-base A.A. calculated by (mol of absorbed CO₂)/(mol of IL), mass-base A.A. (CO₂ mol/absorbent kg), and pressure-reduced A.A. (mol/mol/final partial pressure (bar) of CO₂). In order to evaluate the recyclability of the aqIL-DM ([N₂₂₂₂][Gly] + PAlAm), the sample was regenerated by heating in the infrared humidity meter at 50˚C, 80˚C or 100˚C for 1 hour. The reduced water content was almost recovered by keeping the sample under 100% humidity for 6 hrs.

3. Results and Discussion

3.1. Preparation of DMs

AqIL-DMs were prepared by mixing aqueous solutions of [N₂₂₂₂][Gly] (water content: 55%), [N₂₂₂₂][Ala] (54%) and [N₂₂₂₂][Lys] (48%), with the hydrophobic silica nano particles. The first two ILs provided powdery DMs, while the last one resulted in only a soufflé-like substance. This failure was first ascribed to the lesser amount of water than those of the others. Thus, another [N₂₂₂₂][Lys] sample containing 57% of water was tested but the result was the same; a soufllé-like substance was obtained. Since one lysine molecule has two amino groups, it was expected that the surface of the aqueous [N₂₂₂₂][Lys] solution may be more polar than those of the other two. In fact, however, the long side chain, i.e., four methylene groups, seemed to lower the surface tension of the aqIL.

Aqueous [N₂₂₂₂][Gly] containing 10 wt% of PAlAm (1.8 g of [N₂₂₂₂][Gly], 0.6 g of PAlAm, 3.6 g of water) proved to provide a powdery DM, while only a paste- like D-PS was obtained with the 10 wt% polymer solution. The CO₂ absorption performance of the former DM was investigated.

3.2. CO₂ Absorption Performance

3.2.1. Comparison among the ILs

CO₂ absorption behaviors for the three kinds of aqIL-DMs are shown in Figure 1 as the time course of the partial pressure. The figure also shows the result for the bulk aq. [N₂₂₂₂][Gly] to see how the absorption speed was enhanced by using the aq.IL in the DM form. In the case of the DM, it took ca.5 min to reach 90% of the final absorption level, which was much shorter than that of the bulk; ca.270 min. This absorption acceleration more than 50 times is comparable to that reported in our previous study on DMs prepared with [bmim][ace] [13] . As for [N₂₂₂₂][Lys], it took ca.15 min. This slower absorption than those of DMs prepared from [N₂₂₂₂][Gly] and [N₂₂₂₂][Ala] may be because the [N₂₂₂₂][Lys] DM was not powdery but soufllé-like.

The experimental results on the absorption abilities are summarized on Table 1. As for the absorption ability measured by (mol of absorbed CO₂)/(mol of IL) (A.A. (mol/mol)) given in the 6^th column, it was highest for the [N₂₂₂₂][Lys]DM, ca. two times of those of the other DMs. This means that either of the two amino groups of lysine worked as an equivalent reaction site to accommodate one CO₂ molecule. The A.A. (mol/kg), which was given in the 7^th column, is a measure of CO₂ absorption ability which was calculated by (mol of absorbed CO₂)/(kg of DM or bulk liquid). Since the aqILs contained water of more than 50%, the mass-base A.A. would be useful when one compares CO₂ absorption materials for actual application in industry. In the case of 20% - 30% monoethanolamine (MEA) solution, which is often used to remove CO₂ from natural gas in the industrial scale, the A.A. (mol/kg) is ca. 1.6 - 2.5 [9] . Although the A.A. values obtained for the aqIL-DMs were lower than or at most comparable to the MEA system, the former values may be improved by reducing the water content, as shown in the following section. The A.A. (mol/mol/bar) in the last column is a measure of CO₂ absorption at the same equilibrium partial pressure. Since the present CO₂ absorption mechanism is not physisorption but via the chemical reaction with the amino group, the mol-base A.A. may not be simply proportional to the CO₂ partial pressure at equilibrium. However, the pressure-reduced

*Soufllé-like material.

A.A. may be utilized as a qualitative measure for CO₂ absorption material that is efficient even at a low pressure <1 bar. With this measure, the lysine-based aqIL- DM proved to be more efficient than the others.

3.2.2. Effects of Water Content

It has been known that A.A. (mol/mol) of the amino acid-based ILs may be improved by addition of substantial amount of water. For example, Zhang et al. [16] reported that the mol-base absorption ability of [N₁₁₁₁][Gly] (0.97 bar, 298 K) increased by adding water; from ca.0.17 for pure IL to 0.60 at 70% water. Further, according to Anderson et al. [10] , several amino acid-based ILs containing tetrabutylphosphonium cation ([P₄₄₄₄]) showed maximum CO₂ absorption when the water/IL mol ratio was around 1.5 (water content: ca. 6% - 8%) while the A.A. (mol/mol) of [P₄₄₄₄][Ala] decreased with increasing water content up to water/IL ratio = 2.06, at least. These two studies, which were performed under different water content conditions, suggest that the A.A. of amino acid- based ILs may largely depend on the water content. In fact, the present [N₂₂₂₂] [amino acid] IL systems (aqIL-DMs) also showed similar water-dependency. Figure 2 shows the results for [N₂₂₂₂][Gly] of three different water contents as the time course of the partial pressure of CO₂. The most significant mol-base CO₂ absorption was obtained with the aqIL containing 55% of water (the same plot as that in Figure 1), while the second one and the least one were observed for systems containing 16% and 30% of water, respectively. Similar non-mono- tonic water content dependencies were also observed for [N₂₂₂₂][Ala] and [N₂₂₂₂] [Lys] systems. The results are shown on Table 2. In the alanine-IL systems, the highest A.A. (mol/mol) was obtained in the absence of water, which is contrastive to the other two systems which showed significant CO₂ absorption in the presence of large amounts of water. As for the alanine system, some literature data are available in addition to that by Anderson et al. [10] ; according to Jiang et al. [3] , the mol-base A.A. of the bulk [N₂₂₂₂][Ala] was ca. 0.43˚C at 40˚C and ambient pressure. Although the constituting cation was different, aq. [P₄₄₄₄][Ala] solutions containing 1% or less of water have been reported to absorb almost equimolar CO₂ [2] [10] . The latter results may be compared with the present one for [N₂₂₂₂][Ala] in the absence of water. Since the DMs were treated in air, the

*These row data are the same as those given on Table 1.

0% water sample possibly contained small amount of water after the heating at 80˚C. Although some CO₂ absorption mechanisms of amine-functionalized absorbents in the presence and absence of water have been proposed by different researchers [2] [6] [16] , the detailed examination is out of scope of the present study. We just note here that the CO₂ absorption performance on the mass base, which must be important from an application point of view, was most significant for [N₂₂₂₂][Ala] in the absence of water and [N₂₂₂₂][Lys] in the presence of both large and small amount of water.

3.2.3. PAlAm-Containing AqIL-DM

CO₂ absorption for the powdery aqIL([N₂₂₂₂][Gly])-DM was measured three times. As shown later, each measurement was carried out as a first absorption measurement before the CO₂ desorption by the heat treatment which was done to evaluate the recyclability of the DM. Figure 3 and Table 3 show only a typical

*Average of three measurements (0.70, 0.65 and 0.61).

In order to figure out what caused the higher performance, we first examine the A.A. values of the respective bulk systems. As for the bulk PAlAm solution, Nagai et al. [17] reported that the A.A. (mol/kg) for the 10 wt% solution reached a saturated value, 0.86, after ca.120 min of CO₂ flow through the sample at 25˚C and under the ambient pressure. The A.A. value and the absorption time are both superior to those obtained in the present study; 0.75 mol/kg and ca.500 min. However, the present A.A. value, 0.75, may be comparable to the literature value if the lower pressure condition (p_final = 0.864 bar) was taken into account. Further, the several times longer absorption time may be ascribed to the difference in the CO₂ absorption method. On the other hand, the 0.43 as the mol-base A.A. is significantly less than ca. 0.65 of a typical CO₂ absorbent, monoethanolamine (MEA) obtained at 40˚C under the corresponding pressure (~0.85 bar) [18] [19] . According the proposed mechanism for the CO₂ absorption into aqueous MEA solution,

CO₂ may be absorbed via the zwitterion formation, the carbamate formation and the recovering free amino group via the dissociation of the protonated MEA. Although the above three reactions comprise just a part of the complicated reaction mechanism [18] , they suggest that MEA may absorb CO₂ in a molar ratio larger than 0.5, which is in fact consistent with their experimental observations. Thus, the 0.43 as the mole-base A.A. value obtained for the present aqueous PAlAm solution was an unexpected result because the polymer also contains primary amino groups as MEA does. As a possible cause for the poor result, we note the phase separation and/or the gel formation due to the electrostatic cross linking formation among the carbamate anions and the protonated amino groups, which has been reported by several authors [13] [17] [19] [20] . In fact, the much slower absorption into the bulk polymer solution compared with the bulk aqIL, which is easily seen from the respective time courses of the CO₂ pressure shown in Figure 3, may be ascribed to the gel formation or phase separation due to CO₂ absorption in the former system. Thus, the lower A.A. performance of the PAlAm compared with that of the small amines may be because the gel formation or phase separation partially secluded amino groups in the separated phase, keeping them intact.

This crosslinking reaction of PAlAm via CO₂ absorption should also occur in the bulk IL + PAlAm system, and in fact quite a distinguishing absorption behavior, apparently two-step absorption, was observed. Comparing the first and the second absorption “width” (decrements in the CO₂ partial pressure) with those of the respective bulk systems (the bulk aqIL and the bulk aq. PAlAm), the first one seems to be almost corresponding to the absorption by the IL and followed by the second one, i.e., absorption by the polymer. Needless to say, since both absorption modes should occur simultaneously, the existence of such a marked kink in the CO₂ absorption suggests that the absorption by the polymer was largely retarded in the IL + PAlAm system. As a possible mechanism to effect such retardation, a “preferential reaction” with the IL rather than that with the polymer may be considered. As illustrated in a schema for the CO₂ diffusion through the IL + PAlAm system (Figure 4), most CO₂ molecules may be “trapped” with the IL amino groups before the reaction with the polymer amino groups. Further, CO₂ molecules in the polymer coil, which was depicted as a circle in the figure, would take a much longer time to diffuse through than that through the IL region, because of the crosslinking produced due to the reaction with CO₂. Thus, it seems that in the first step (t < ca.1000 min), PAlAm, which was near the surface of the bulk IL + PAlAm solution, could only react with CO₂, while most IL molecules reacted with CO₂ rather freely, and after the first step period, free access of CO₂ to the other polymer amino groups became substantial.

Are the above speculations on the bulk systems consistent with the absorption behavior of the DM consisting of the IL and PAlAm? First of all, one must note that the time course profile of the CO₂ partial pressure was smooth and no kink was observed. This may be because the maximum diffusion length of CO₂ in the

liquid phase to reach the innermost amino groups from the surface was much shorter than those in the respective bulk systems; order of 100 μm for the former [12] and ~1 cm for the latter. Thus, the reaction of CO₂ with the respective amino groups, i.e., IL and PAlAm, could almost simultaneously proceed in the DM system. In fact, the higher performance of the DM compared with both bulk systems may also be explained by the same reasoning. Namely, the small size (~100 μm) of the DM droplets enabled quick access of CO₂ to all the amino groups before the gel formation effectively retarded the CO₂ diffusion. Then, the total A.A. in molar base should become higher than that of the bulk system, in which the polymer amino groups were partially inaccessible to CO₂ due to the gel formation. Thus, the present experimental results strongly suggest that the shortcoming involved in the bulk (IL + PAlAm) system may be overcome by employing the same system in the DM form.

As for the mass-base A.A., Table 3 tells us the DM was much superior to either of the bulk systems. This comes from the efficient use of water; although water was necessary to improve the CO₂ absorption ability of [N₂₂₂₂][Gly], water itself only scarcely absorb the gas. Thus, the large amount of water in the DM just reduced the mass-base A.A. However, the incorporation of the PAlAm into the DM could significantly improve the mass-base A.A. (1.9 vs. 0.95) because the water phase including PAlAm could also work as a substantial CO₂ absorbent.

3.3. Recyclability of the DM

Recyclability is one of the prerequisites that should be met by CO₂ absorbents for industrial application. In the case of CO₂ capture by aq. MEA [10] [21] , the regeneration of the absorbent has been performed by stripping with water vapor at 100˚C - 120˚C. In the present study, CO₂ removal was carried out simply by heating CO₂-saturated aqIL ([N₂₂₂₂][Gly])-DM samples, which contained 10% PAlA maq. solution, at 50˚C, 80˚C or 100˚C. Then, the regenerated materials were subject to the second CO₂ absorption, and the ratio of the resultant A.A. in the molar base to that of the original one obtained before the first heat treatment was used as a measure for the recyclability. The results are summarized on Table 4. Unfortunately all the trials to recycle the DM material proved to be unsuccessful; although the recovery of water content was satisfying to some extent, the A.A. recovery was far less from ca. 82% of aq. MEA solution which was regenerated at 120˚C [10] . As for the reason for the miserably bad results, it may be appropriate to refer to the study by Nagai et al. again [17] . The authors successfully prepared PAlAm hydrogel by crosslinking the polymer amino groups via urea bond under high pressure of CO₂ (3.5 MPa) at 170˚C. The same reaction might occur in the DM particles by the heat treatment. Namely, many free amino groups of the polymer were consumed by the irreversible urea bond formation and the resultant gel formation also made the free IL inaccessible to CO₂. In fact, a heat treatment at 50˚C for 10 wt% PAlAm solution resulted in a glassy substance as shown in Figure 5 and the solid was swollen but insoluble in water. This observation strongly suggested that some irreversible crosslinking via cova-

*Same data as those given on Table 3.

lent bond formation occurred between the polymer chains. Thus, in order to apply the PAlAm-containing aqIL-DM for industrial use, the irreversible urea group formation due to the heat treatment must be inhibited.

4. Conclusions

In the present study we examined the CO₂ absorption performance of several kinds of DMs to find the following results:

1) Powdery DMs were successfully prepared with amino acid-based IL/water mixtures, i.e., [N₂₂₂₂][Gly], [N₂₂₂₂][Ala] as aqueous solutions of ca. 50%, while only soufflé type of DM was obtained for [N₂₂₂₂][Lys] even when the water content was as high as ca.60%;

2) The mol-base CO₂ absorption ability was largely dependent on the water content and the IL species;

3) Whereas only a paste-like D-PS was obtained with the 10 wt% PAlAm solution, powdery DM was successfully prepared with [N₂₂₂₂][Gly] containing 10 wt% of the polymer and 60 wt% of water;

4) The aqIL-DM ([N₂₂₂₂][Gly] + PAlAm) showed higher CO₂ absorption abilities (especially the mass-base A.A.) than those of DM containing only the IL;

5) The recyclability of the PAlAm-containing aqIL-DM proved to be rather poor, probably because the heat treatment, which was necessary to remove the absorbed CO₂ from the absorbent, irreversibly crosslinked the polymer, leading to gelation of the inner phase of the DM particles.

Thus, the present study demonstrated that the amino acid-based aqIL-DM containing PAlAm is a promising absorbent material for CO₂. In order to overcome the shortcoming found for the recyclability, it may be effective to use a Dry Gel system in which PAlAm constitutes the gel phase [17] swollen with the aqIL. The results will be soon reported in a forthcoming paper.

Acknowledgements

This work was supported by JSPS KAKENHI grant number JP15K05584.

Compliance with Ethical Standards

Conflict of interest the authors declare that they have no conflict of interest.